Explore global development with R

Today, you will load a filtered gapminder dataset - with a subset of data on global development from 1952 - 2007 in increments of 5 years - to capture the period between the Second World War and the Global Financial Crisis.

Your task: Explore the data and visualise it in both static and animated ways, providing answers and solutions to 7 questions/tasks below.

Get the necessary packages

First, start with installing the relevant packages ‘tidyverse’, ‘gganimate’, and ‘gapminder’.

Look at the data and tackle the tasks

First, see which specific years are actually represented in the dataset and what variables are being recorded for each country. Note that when you run the cell below, Rmarkdown will give you two results - one for each line - that you can flip between.

str(gapminder)
## tibble [1,704 × 6] (S3: tbl_df/tbl/data.frame)
##  $ country  : Factor w/ 142 levels "Afghanistan",..: 1 1 1 1 1 1 1 1 1 1 ...
##  $ continent: Factor w/ 5 levels "Africa","Americas",..: 3 3 3 3 3 3 3 3 3 3 ...
##  $ year     : int [1:1704] 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 ...
##  $ lifeExp  : num [1:1704] 28.8 30.3 32 34 36.1 ...
##  $ pop      : int [1:1704] 8425333 9240934 10267083 11537966 13079460 14880372 12881816 13867957 16317921 22227415 ...
##  $ gdpPercap: num [1:1704] 779 821 853 836 740 ...
unique(gapminder$year)
##  [1] 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 2002 2007
head(gapminder)
## # A tibble: 6 × 6
##   country     continent  year lifeExp      pop gdpPercap
##   <fct>       <fct>     <int>   <dbl>    <int>     <dbl>
## 1 Afghanistan Asia       1952    28.8  8425333      779.
## 2 Afghanistan Asia       1957    30.3  9240934      821.
## 3 Afghanistan Asia       1962    32.0 10267083      853.
## 4 Afghanistan Asia       1967    34.0 11537966      836.
## 5 Afghanistan Asia       1972    36.1 13079460      740.
## 6 Afghanistan Asia       1977    38.4 14880372      786.

The dataset contains information on each country in the sampled year, its continent, life expectancy, population, and GDP per capita.

Let’s plot all the countries in 1952.

theme_set(theme_bw())  # set theme to white background for better visibility

ggplot(subset(gapminder, year == 1952), aes(gdpPercap, lifeExp, size = pop)) +
  geom_point() +
  scale_x_log10()

We see an interesting spread with an outlier to the right. Answer the following questions, please:

  1. Why does it make sense to have a log10 scale on x axis?

Because the distribution is a growth function, by taking it on a log scale, every step up the y-axis is an identical percent change in real GDP per capita. This makes the graph easier interpretable.

  1. Who is the outlier (the richest country in 1952 - far right on x axis)?

The outlier, or the richest country in 1952 is kuwait:

gapminder %>% subset(.$year == 1952) %>% arrange(desc(gdpPercap))
## # A tibble: 142 × 6
##    country        continent  year lifeExp       pop gdpPercap
##    <fct>          <fct>     <int>   <dbl>     <int>     <dbl>
##  1 Kuwait         Asia       1952    55.6    160000   108382.
##  2 Switzerland    Europe     1952    69.6   4815000    14734.
##  3 United States  Americas   1952    68.4 157553000    13990.
##  4 Canada         Americas   1952    68.8  14785584    11367.
##  5 New Zealand    Oceania    1952    69.4   1994794    10557.
##  6 Norway         Europe     1952    72.7   3327728    10095.
##  7 Australia      Oceania    1952    69.1   8691212    10040.
##  8 United Kingdom Europe     1952    69.2  50430000     9980.
##  9 Bahrain        Asia       1952    50.9    120447     9867.
## 10 Denmark        Europe     1952    70.8   4334000     9692.
## # … with 132 more rows

Next, you can generate a similar plot for 2007 and compare the differences

ggplot(subset(gapminder, year == 2007), aes(gdpPercap, lifeExp, size = pop)) +
  geom_point() +
  scale_x_log10()

The black bubbles are a bit hard to read, the comparison would be easier with a bit more visual differentiation.

Tasks:

  1. Differentiate the continents by color, and fix the axis labels and units to be more legible (Hint: the 2.50e+08 is so called “scientific notation”, which you might want to eliminate)
options(scipen = 999)
ggplot(gapminder %>% subset(.$year == 2007), aes(gdpPercap, lifeExp, size = pop, col = continent)) +
  geom_point() +
  scale_x_log10()

  1. What are the five richest countries in the world in 2007?
gapminder %>% subset(.$year == 2007) %>% arrange(desc(gdpPercap)) %>% top_n(., 5)
## Selecting by gdpPercap
## # A tibble: 5 × 6
##   country       continent  year lifeExp       pop gdpPercap
##   <fct>         <fct>     <int>   <dbl>     <int>     <dbl>
## 1 Norway        Europe     2007    80.2   4627926    49357.
## 2 Kuwait        Asia       2007    77.6   2505559    47307.
## 3 Singapore     Asia       2007    80.0   4553009    47143.
## 4 United States Americas   2007    78.2 301139947    42952.
## 5 Ireland       Europe     2007    78.9   4109086    40676.

As you can see the 5 richest countries in 2007 are Norway, Kuwait, Singapore, United States and Ireland.

Make it move!

The comparison would be easier if we had the two graphs together, animated. We have a lovely tool in R to do this: the gganimate package. Beware that there may be other packages your operating system needs in order to glue interim images into an animation or video. Read the messages when installing the package.

Also, there are two ways of animating the gapminder ggplot.

Option 1: Animate using transition_states()

The first step is to create the object-to-be-animated

anim <- ggplot(gapminder, aes(gdpPercap, lifeExp, size = pop)) +
  geom_point() +
  scale_x_log10()  # convert x to log scale
anim

p <- ggplot(
  gapminder, 
  aes(x = gdpPercap, y=lifeExp, size = pop, colour = country)
) +
  geom_point(show.legend = FALSE, alpha = 0.7) +
  scale_color_viridis_d() +
  scale_size(range = c(2, 12)) +
  scale_x_log10() +
  labs(x = "GDP per capita", y = "Life expectancy")
#> Error in ggplot(gapminder, aes(x = gdpPercap, y = lifeExp, size = pop, : could not find function "ggplot"
p

#> Error in eval(expr, envir, enclos): object 'p' not found
p + transition_time(year) +
  labs(title = "Year: {frame_time}")

#> Error in eval(expr, envir, enclos): object 'p' not found
file_renderer(dir = ".", prefix = "gganim_plot", overwrite = FALSE)
## function (frames, fps) 
## {
##     if (!dir.exists(dir)) 
##         dir.create(dir, showWarnings = FALSE, recursive = TRUE)
##     new_names <- file.path(dir, sub("gganim_plot", prefix, basename(frames)))
##     if (any(!file.copy(frames, new_names, overwrite = overwrite))) {
##         warning("file_renderer failed to copy frames to the destination directory", 
##             call. = FALSE)
##     }
##     invisible(new_names)
## }
## <bytecode: 0x7f93fb4346f0>
## <environment: 0x7f93fb42f928>

This plot collates all the points across time. The next step is to split it into years and animate it. This may take some time, depending on the processing power of your computer (and other things you are asking it to do). Beware that the animation might appear in the bottom right ‘Viewer’ pane, not in this rmd preview. You need to knit the document to get the visual inside an html file.

anim + transition_states(year, 
                      transition_length = 1,
                      state_length = 1)

…

Notice how the animation moves jerkily, ‘jumping’ from one year to the next 12 times in total. This is a bit clunky, which is why it’s good we have another option.

Option 2 Animate using transition_time()

This option smoothes the transition between different ‘frames’, because it interpolates and adds transitional years where there are gaps in the timeseries data.

gapminder = gapminder
anim2 <- ggplot(gapminder, aes(gdpPercap, lifeExp, size = pop)) +
  geom_point() +
  scale_x_log10() + # convert x to log scale
  transition_time(year) +
  labs(x = "GDP per capita", y = "Life expectancy")

animate(anim2, renderer = gifski_renderer())

The much smoother movement in Option 2 will be much more noticeable if you add a title to the chart, that will page through the years corresponding to each frame.

Now, choose one of the animation options and get it to work. You may need to troubleshoot your installation of gganimate and other packages

  1. Can you add a title to one or both of the animations above that will change in sync with the animation? (Hint: search labeling for transition_states() and transition_time() functions respectively)

  2. Can you made the axes’ labels and units more readable? Consider expanding the abreviated lables as well as the scientific notation in the legend and x axis to whole numbers.

pacman::p_load(viridis)

anim3 <- ggplot(gapminder, aes(gdpPercap, lifeExp, size = pop, col = continent)) +
  geom_point() +
  scale_x_log10() + # convert x to log scale
  labs(subtitle = "{closest_state}") + # add frame variable as subtitle
  transition_states(states = year) + 
  labs(x = "GDP per capita", y = "Life expectancy")
anim3

animate(anim3, renderer = gifski_renderer())

gapminder_unfiltered = gapminder_unfiltered
  1. Come up with a question you want to answer using the gapminder data and write it down. Then, create a data visualisation that answers the question and explain how your visualization answers the question. (Example: you wish to see what was mean life expectancy across the continents in the year you were born versus your parents’ birth years). [Hint: if you wish to have more data than is in the filtered gapminder, you can load either the gapminder_unfiltered dataset and download more at https://www.gapminder.org/data/ ]

I expect there to be a different dependency between population number and gdp in Asia, europe and america. Therefore, I run a linear model with an interaction between GDP and continent, then I filter out only these three continents:

lm1 = lm(pop~gdpPercap*continent, data = gapminder_unfiltered %>% filter(.$continent %in% c("Europe", "Asia", "Americas")))
interact_plot(lm1, pred = gdpPercap, modx = continent)

Here we can see that there is a positive relationship between population size and GDP in America, while it is negative in Asia. Further, there is no relationship between population size and GDP in europe. This could speak to the different political stances in the different continents.

To understand this realtionship with a time dependency, I use gganimate, to visualize the relationship in time:

anim4 <- ggplot(gapminder %>% filter(.$continent %in% c("Europe", "Asia", "Americas")) %>% mutate(pop = as.numeric(pop)), aes(x = gdpPercap, y = pop, col = continent)) +
  geom_point() +
  scale_x_log10() + # convert x to log scale
  labs(subtitle = "{closest_state}") + # add frame variable as subtitle
  transition_states(states = year) + 
  labs(x = "GDP per capita", y = "Population Size") + 
  facet_wrap(~continent)
anim4
animate(anim4, renderer = gifski_renderer())